[PATCH 05/11] rcu: clean all rcu_read_unlock_special after report qs

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



rcu_preempt_deferred_qs_irqrestore() is always called when
->rcu_read_lock_nesting <= 0 and there is nothing to prevent it
from reporting qs if needed which means ->rcu_read_unlock_special
is better to be clearred after the function. But in some cases,
it leaves exp_hint (for example, after rcu_note_context_switch()),
which is harmess since it is just a hint, but it may also intruduce
unneeded rcu_read_unlock_special() later.

The new code write to exp_hint without WRITE_ONCE() since the
writing is protected by irq.

Signed-off-by: Lai Jiangshan <laijs@xxxxxxxxxxxxxxxxx>
---
 kernel/rcu/tree_plugin.h | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/kernel/rcu/tree_plugin.h b/kernel/rcu/tree_plugin.h
index 9fe8138ed3c3..bb3bcdb5c9b8 100644
--- a/kernel/rcu/tree_plugin.h
+++ b/kernel/rcu/tree_plugin.h
@@ -439,6 +439,7 @@ rcu_preempt_deferred_qs_irqrestore(struct task_struct *t, unsigned long flags)
 	 * t->rcu_read_unlock_special cannot change.
 	 */
 	special = t->rcu_read_unlock_special;
+	t->rcu_read_unlock_special.b.exp_hint = false;
 	t->rcu_read_unlock_special.b.deferred_qs = false;
 	if (special.b.need_qs) {
 		rcu_qs();
@@ -626,7 +627,6 @@ static void rcu_read_unlock_special(struct task_struct *t)
 		local_irq_restore(flags);
 		return;
 	}
-	WRITE_ONCE(t->rcu_read_unlock_special.b.exp_hint, false);
 	rcu_preempt_deferred_qs_irqrestore(t, flags);
 }
 
-- 
2.20.1




[Index of Archives]     [Linux Samsung SoC]     [Linux Rockchip SoC]     [Linux Actions SoC]     [Linux for Synopsys ARC Processors]     [Linux NFS]     [Linux NILFS]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]


  Powered by Linux